You are here
[LS-DYNA] Attenuating vibrations of shell element after an explosion
Fri, 2010-02-26 09:32 - jgrzmmm
Hello!
The model I'm working on in LS-DYNA is a simple pyramid-shaped shell constrained along the bottom edges with all DOF taken. The material used is MAT_PIECEWISE_LINEAR_PLASTICITY. Above the model there's a LOAD_BLAST (Air burst) placed, which deforms the model after the explosion.
The issue is that the model seems to vibrate ininitely after the explosion, what can be seen on the attached images (first one shows the graph of "Rigid Body Displacement", second one shows close-up of the line within the yellow rectangle).
I've already tried various HOURGLASS options including changing linear bulk viscosity coefficient, but had no improvement. What can be done to attenuate those vibrations?
Best regards,
Julian
Forums:
Free Tags:
![Subscribe to Comments for "[LS-DYNA] Attenuating vibrations of shell element after an explosion"](https://imechanica.org/misc/feed.png){kind=small}
My thoughts
Hello Julian!
I have some expertise on modeling blast loads, so please let me share some thoughts that came into my mind according to your question.
1. First I do not know what is your goal with the simulation, but my experience shows, that the important things happen in the first 10-25 ms when the response to blast loading is in focus. Your simulation shows the same pattern, so the time interval you concern is irrelevant (unless you investigate the rigid body motion - e.g. free falling - of the system)
2. The vibration you show is five orders of magnitude smaller than your maximal displacement. This is far less than any experimental uncertainity. If it would be 100 times bigger, it still would be within the engineering accuracy.
3. I think your problem is not connected to hourglassing rather a residual elastic vibration of the system, which leads directly to my next point.
4. Most materials in explicit FE codes do not contain material damping. This is no harm, since modeling blast loads take time in a few milliseconds where structural damping is irrelevant. The only thing that consumes energy is plasticity. All the plastic response occurs in the first 25 ms - according to your graph. As the plastic deformation takes place, your velocities - hence the force to deform your system - will get smaller and you fall back to the elastic region. From thet time you will have some elastic vibration in the system without anything to absorb the energy. It will stay like this forever.
5. Not to mention the fact that explicit codes ALWAYS have a vibration in them, since they do not solve the algebraic equations directly. Just build in a single element without any load or BC and let it run for 10-100 ms. You will get stresses at about 1e-5.
My advice would be, just simply forget the vibration and maybe as a result give an average value of the last 75 ms.
I hope I had some good ideas for your future analysis.
Regards, Andras
Thank you for your response
Thank you for your response Andras!
The results I need are the peak transient deflection and the stabilized response, which can be actually seen for the first 25ms.
It's true that the peak-to-peak amplitude is only about 0.00001 compared to the average residual deflection of 0.02426, so I will use this result in my report.
I was mainly concerned about choosing the right material and had doubts, that maybe the one I used, was causing such infinite vibrations. Your answer convinced me that it wasn't the main "problem".
I decided that I need only first 25-30ms of the simulation, it will surely save a lot of computational time!
Thank you for your help Andras, gave me a great insight into some matters I deal with in my simulations!
Best regards,
Julian